We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
For the 40-year-old Autodesk, known for its design and creation software (including AutoCAD) used by professionals in industries including architecture, engineering, construction, manufacturing and entertainment, artificial intelligence (AI) has become a must to help boost creativity and collaboration.
“A common theme is helping the designer,” said Tonya Custis, director of artificial intelligence research at Autodesk, whose team includes fifteen AI research scientists based in San Francisco, Toronto and London.
But AI will also help Autodesk expand in the metaverse. According to Custis, Autodesk’s use of artificial intelligence is also helping to tackle challenges around “geometry understanding” – to help contextualize the geometric world around us – which will be “super-important” as the metaverse expands, in terms of speeding up animation and CGI processes as well as in architecture and engineering.
“It’s about how we can understand the geometry of the world around us – not just of objects, but of space,” she said, adding that Autodesk’s AI efforts will “absolutely” be important as the metaverse evolves. “For example, how is a space organized? What are the things in it? How can we break it down into geometry and, then, what are its functions – because a computer does not know that.”
AI investments democratize technology
Media coverage acknowledges that Autodesk, along with companies such as Meta, Roblox, Microsoft and Nvidia, may play a role in building the metaverse.
That may include the role played by Autodesk’s investments and acquisitions: The San Rafael, California-based company recently announced an investment in Radical, a New York-based developer whose proprietary AI combines modern deep learning strategies, human biomechanics, and computer graphics to estimate, track, and reproduce skeletal joint rotations in 3D from a single conventional video feed. From videos to metaverses, this data can be used to automate the animation of 3D characters and avatars – and requires no special hardware, training, or custom coding.
The investment in Radical follows Autodesk’s acquisition of Moxion, with its cloud solution for digital dailies, in January and last November’s acquisition of cloud-based animation pipeline software firm Tangent Labs.
“Autodesk has a lot of tools that people use to make things in the professional space of things like animation and movies, but as far as content creation goes, these tools are becoming more ubiquitous,” Custis explained. “So Autodesk’s investment in a company like Radical democratizes a lot of that technology.”
Autodesk’s AI to help, not hinder
But Autodesk is most famous for its work in the architecture, engineering and construction space, particularly through their AutoCAD software. “My AI research team, in particular, works on things like floor plan generation, while there are some projects product teams are working on using machine learning to make command sequences easier, to make it easier to import information from drawing,” she said. “A lot of architects like to use paper to do their designs and then they have to be translated into CAD – so that’s a real waste of time for them.”
Since many AutoCAD users are experts – often even getting graduate degrees in the use of the software – there’s a fine line between automation that is helpful and taking control away. “It’s a lot about how we provide algorithms that automate things that make sense that will save them time, but also giving them the agency to make choices, or give them recommendations that they can then choose,” she said. “It’s definitely a collaborative AI environment on the AEC side.”
For manufacturers, Custis said her team works a great deal with Autodesk’s Fusion product, on issues such as deep learning for 3D CAD models. “For example, we teach the computer to learn how to put assemblies together, such as all the parts you need to build a unicycle,” she said. “And then, can we teach specific robots to do that, once we understand what the steps are, what’s required, how the pieces go together?”
AI and generative design
Autodesk is also highly-focused on artificial intelligence-based generative design, in which “designers or engineers input design goals into the generative design software, along with parameters such as performance or spatial requirements, materials, manufacturing methods, and cost constraints. The software explores all the possible permutations of a solution, quickly generating design alternatives. It tests and learns from each iteration what works and what doesn’t.”
While debate around the use of large language models is all the rage at the moment, they offer use cases that are very relevant to Autodesk, especially in media and entertainment, said Custis.
“It’s definitely something we’re looking at closely and we’re actually also working with OpenAI,” she said. “I think generative models are really exciting in our space – the trajectory in machine learning is usually first we do stuff on text, then we do stuff on pictures, then we do stuff on videos, then we do stuff in 3D – so all of this is happening right now.”
The ultimate goal at Autodesk, she reiterated, is to use AI to help users have more time to be more creative.
“We don’t want to replace them, we don’t want to take their job from them,” she said. “But we do want to give them more flexibility and agency about how they use their time and support that creativity.”
As for Autodesk’s impact on the metaverse, Custis said the future remains to be seen.
“There’s a place there and a lot of the work my team is working on in AI research is pretty applicable,” she said. “But I can’t speculate how those particular things will play out.”